Deep Learning-based Automatic Lung Segmentation on Multiresolution CT Scans from Healthy and Fibrotic Lungs in Mice.

To develop a model to accurately segment mouse lungs with varying levels of fibrosis and investigate its applicability to mouse images with different resolutions. In this experimental retrospective study, a U-Net was trained to automatically segment lungs on mouse CT images. The model was trained ( The trained model yielded a high median DSC in both test datasets: 0.984 (interquartile range [IQR], 0.977-0.988) in group A and 0.966 (IQR, 0.955-0.972) in group B. The median HD in both test datasets was 0.47 mm (IQR, 0-0.51 mm [group A]) and 0.31 mm (IQR, 0.30-0.32 mm [group B]). Spatially resolved quantification of differences toward reference masks revealed two hot spots close to the air-tissue interfaces, which are particularly prone to deviation. Finally, for the higher-resolution mouse CT images, the median DSC was 0.905 (IQR, 0.902-0.929) and the median 95th percentile of the HD was 0.33 mm (IQR, 2.61-2.78 mm). The developed deep learning-based method for mouse lung segmentation performed well independently of disease state (healthy, fibrotic, emphysematous lungs) and CT resolution.

Informations de copyright

2022 by the Radiological Society of North America, Inc.

Déclaration de conflit d'intérêts

Disclosures of Conflicts of Interest: F.S. No relevant relationships. P.S. No relevant relationships. M.M. No relevant relationships. C.Z. No relevant relationships. C.S. No relevant relationships. K.R. No relevant relationships. N.B. No relevant relationships. A.K. No relevant relationships. H.W. Grant from Dutch Cancer Society (KWF Kankerbestrijding) to author’s institution; minority shares in Radiomics (Liege, Belgium). L.D. No relevant relationships. P.L. Member of the external Scientific Advisory Board at CRUK Lung Cancer Center of Excellence; member of the Scientific Advisory Board (CSA) at Convert Pharmaceuticals; Chief Scientific Advisor at Oncoradiomics; member of the Advisory Board at Benelux Health Ventures SPRL; Chief Scientific Advisor (CSA), member of the SAB at The Medical Cloud Company (MedC2); co-founder and shareholder (minority) at Oncoradiomics; co-founder and shareholder (minority) at Convert Pharmaceuticals; shareholder (minority) at The Medical Cloud Company (Med2C); shareholder (minority) at LivingMed Biotech; member of the Scientific Advisory Board at The Institute of Cancer Research; Scientific Advisor at Institut Jules Bordet, ULB; Scientific Advisor at Medneon; shareholder (minority) at Exomnis Biotech; scientific review of grants at Cancer Research UK Review Panel; consultant at Bristol Myers Squibb; Chief Scientific Advisor at Exomnis Biotech. J.D. Grants/contracts from/with Viewray, CRI – The Clinical Research Institute, Accuray International Sarl, RaySearch Laboratories, Vision RT, Merck Serono, Astellas Pharma, AstraZeneca, Siemens Healthcare, Solution Akademie, Ergomed PLC Surrey Research Park, Quintiles, Pharmaceutical Research Associates, Boehringer Ingelheim Pharma & CoKG, PTW-Freiburg Dr. Pychlau, Nanobiotix, Accuray, Varian; participation on a data safety monitoring board or advisory board for Merck Serono. A.A. Predict MarieCurie innovative training network (ITN), in frame of Horizon 2020 from the European Union, Marie Skłodowska-Curie grant agreement No 766276. M.K. No relevant relationships.